EpilepsyLive ✉ Suggest a correction
Demonstration and educational project by Danielle Boyce. Not medical, legal, or regulatory advice. Sample and template language is provided for illustration and must be reviewed and adapted before any real use. Not affiliated with, endorsed by, or representing any advocacy group, registry, company, or institution named.
Module 7 of 19Part 2 · Data Architecture

HPO, GA4GH & Phenopackets

Goal: Understand the Human Phenotype Ontology (HPO), the GA4GH data sharing framework, and Phenopackets, the emerging standard for exchanging participant phenotype and genomic data.

The Human Phenotype Ontology (HPO)

What is HPO?

The Human Phenotype Ontology is a standardized vocabulary of over 18,000 terms describing human phenotypic abnormalities, symptoms, clinical findings, and disease features. Each term has a unique identifier (e.g., HP:0001250 = Seizure) and is organized in a hierarchical structure.

HPO was developed by the Monarch Initiative and is now the international standard for phenotype description in rare disease research.

hpo.jax.org

Why your registry needs HPO

Without HPO, "seizure" in your registry and "convulsion" in a collaborator's registry may or may not mean the same thing. HPO maps both terms to HP:0001250, making them computationally equivalent.

This is very important for:

  • Genotype phenotype analysis, Linking specific genetic variants to specific clinical features
  • Cross-registry comparison, Comparing phenotype frequencies across institutions
  • Diagnosis support, Tools like Phenomizer and LIRICAL use HPO terms to suggest diagnoses
  • GA4GH / Phenopackets, HPO is the phenotype backbone of the Phenopackets standard (see below)

Implementing HPO in your registry

  1. Map each clinical feature you collect to the most specific applicable HPO term
  2. For each feature, record: HPO term ID, observation status (present/absent/unknown), age at onset, severity modifier
  3. Use HPO's own browser to identify appropriate terms: hpo.jax.org
  4. Avoid creating custom phenotype terms, if the concept exists in HPO, use it
  • OMIM (Online Mendelian Inheritance in Man): Gene-disease relationships, omim.org
  • HGNC (HUGO Gene Nomenclature Committee): Standardized gene symbols, genenames.org
  • ClinVar: Variant-disease assertions, ncbi.nlm.nih.gov/clinvar

The Global Alliance for Genomics and Health (GA4GH)

What is GA4GH?

GA4GH is an international standards-setting body that develops frameworks and technical standards for responsible genomic and health data sharing. It is the closest thing the field has to a global governance authority for genomic data.

ga4gh.org

The GA4GH Framework for Responsible Sharing

The GA4GH framework, adopted by hundreds of institutions globally, establishes principles for:

  • Data access: Who can access genomic data and under what conditions
  • Data security: Technical standards for protecting genomic data in transit and at rest
  • Ethics: Consent standards, benefit sharing, vulnerable populations
  • Identity management: How individuals are identified across systems

Even if your registry doesn't immediately involve genomics, aligning with the GA4GH framework builds credibility with academic and industry partners who operate within it.

Key GA4GH Standards relevant to registries

Standard Purpose
Phenopackets Structured exchange of phenotype + genomic data
Beacon API Query whether a variant or phenotype exists in a dataset
Data Connect Federated search across multiple datasets
Passport/Visas Identity and access management for data sharing
VRS (Variant Representation Spec) Standardized genomic variant description

Phenopackets

What is a Phenopacket?

A Phenopacket is a structured, computable representation of a participant's phenotype, medical history, and genomic data. Think of it as a standardized medical summary designed for machines to read and exchange, a participant's clinical story in a format that any compliant system can interpret.

The Phenopacket schema (v2) includes:

  • Subject: Individual ID, date of birth, sex, taxonomy
  • Phenotypic features: HPO coded observations with onset, severity, and status (present/absent)
  • Diseases: OMIM/ORDO-coded diagnoses with onset
  • Measurements: LOINC-coded lab values and clinical measurements
  • Genomic interpretations: Variant data with ACMG classifications
  • Medical actions: Treatments, procedures, doses
  • Family history: Pedigree information

github.com/phenopackets/phenopacket-schema

Why Phenopackets matter for your registry

Phenopackets are becoming the submission standard for:

  • NCBI ClinVar, Variant submissions increasingly require Phenopacket format
  • Matchmaker Exchange, Connecting undiagnosed participants globally
  • ERDERA, European rare disease data sharing networks
  • Monarch Initiative, Cross-species phenotype analysis
  • GA4GH Beacon, Federated data discovery

Designing your registry to export Phenopackets from the start means your data is immediately accessible to this entire ecosystem.

Getting started with Phenopackets

  1. Review the Phenopackets v2 schema documentation: phenopackets.readthedocs.io
  2. Use the Python or Java SDK for programmatic creation
  3. Validate Phenopackets with the official validator before submission
  4. Map your existing data elements to Phenopacket fields

Checklist

  • [ ] HPO terms mapped to all phenotypic features in registry
  • [ ] HPO term IDs stored alongside display labels (not just text)
  • [ ] OMIM and/or ORDO codes assigned to diagnoses
  • [ ] HGNC gene symbols used for all genetic data
  • [ ] Reviewed GA4GH Framework for data sharing alignment
  • [ ] Evaluated whether Phenopacket export is feasible with chosen platform
  • [ ] Consent language covers genomic data sharing via GA4GH-compliant networks

Key resources

← Module 6 | Module 8: OMOP & Data Standards →